A Finger Forgery Attack Detection Method Based on OCT Volume Data

Abstract

A method for detecting finger forgery attacks based on OCT volume data, comprising: detecting the number of minutiae of internal and external fingerprints and the number of subcutaneous sweat glands; setting the minutiae number thresholds num1, num2, the sweat gland number threshold num3, the internal and external fingerprint matching score threshold t, and the coincidence rate n between the position of subcutaneous sweat glands and the internal fingerprint ridge lines; if the number of internal fingerprint minutiae is more than num1 and the number of external fingerprint minutiae is more than num2, then calculate the internal and external fingerprint matching score; if the number of internal fingerprint minutiae is less than num1, directly determine it as a finger forgery attack; if the number of sweat glands is less than num3, directly determine it as a finger forgery attack; if the number of sweat glands is more than num3, if the internal and external fingerprint matching score is higher than t, the system directly passes the detection; if the number of sweat glands is less than num3 and at the same time the number of internal fingerprint minutiae is less than num1, enter the next step; calculate the coincidence rate between the position of subcutaneous sweat glands and the internal fingerprint ridge lines; if the coincidence rate is higher than the set value n, pass the detection, otherwise determine it as a finger forgery attack.

Description

Technical Field

[0001] The present invention relates to the technical field of biometric forgery attack detection, and in particular to a method for detecting forgery attacks based on OCT volume data according to various modal information inside a finger. Background Art

[0002] Biometrics refers to the inherent physiological characteristics of the human body, which are generally considered to be unique and permanent. Currently, widely used biometrics include, but are not limited to, fingerprints, palm prints, irises, human faces, etc. Biometric recognition refers to automatically identifying an individual based on the physical or behavioral characteristics of the individual. Although many performance evaluations and optimizations have been carried out on biometric technologies, the technology is still vulnerable to forgery attacks. In the increasingly complex use of biometric technologies, fingerprints, as the most widely used biometric, the security issues of its recognition have attracted considerable attention.

[0003] In recent years, applying the internal information obtained by Optical Coherence Tomography (OCT) to fingerprint recognition is a new direction to solve the inherent defects of ordinary commercial fingerprint scanners. As an emerging optical technology, OCT has developed rapidly in recent years with its advantages of real-time, 3D, high sensitivity, label-free, etc. Through the three-dimensional volume data of the finger obtained by OCT, a variety of different but complementary biological modalities can be obtained, and these modal information can be used to implement an efficient forgery attack detection method. Summary of the Invention

[0004] In order to overcome the defects of the existing commercial fingerprint acquisition system with poor security and inability to accurately resist forgery attacks, the present invention provides a finger forgery attack detection method based on OCT volume data with good anti-counterfeiting performance, easy to implement, and applicable to different OCT fingerprint acquisition systems.

[0005] The technical solution adopted by the present invention to solve its technical problems is:

[0006] A finger forgery attack detection method based on OCT volume data, the method comprising the following steps:

[0007] 1) Use an OCT fingerprint acquisition device to obtain the OCT volume data of a finger;

[0008] 2) Obtain an external fingerprint image, an internal fingerprint image, and a subcutaneous sweat gland image respectively;

[0009] 3) Detect the number of minutiae of the internal and external fingerprints and the number of subcutaneous sweat glands respectively;

[0010] 4) Set the thresholds for the number of minutiae num1, num2, the threshold for the number of sweat glands num3, the threshold t for the matching score of the inner and outer fingerprints, and the coincidence rate n between the position of the subcutaneous sweat glands and the inner fingerprint ridges;

[0011] 5) If the number of inner fingerprint minutiae is more than num1 and the number of outer fingerprint minutiae is more than num2, calculate the matching score of the inner and outer fingerprints through a fingerprint matcher; if the number of inner fingerprint minutiae is less than num1, directly determine it as a finger forgery attack.

[0012] 6) If the number of sweat glands is less than num3, directly determine it as a finger forgery attack;

[0013] 7) If the number of sweat glands is more than num3, determine whether the matching score of the inner and outer fingerprints obtained in step 5) is higher than the set threshold t. If it is higher than the threshold t, the system directly passes the detection. If it is lower than the threshold t, go to step 9);

[0014] 8) If the number of sweat glands is less than num3 and at the same time the number of inner fingerprint minutiae is less than num1, go to step 9);

[0015] 9) Locate the position of the subcutaneous sweat glands and calculate the coincidence rate between the position of the subcutaneous sweat glands and the inner fingerprint ridges;

[0016] 10) If the coincidence rate is higher than the set value n, pass the detection; otherwise, determine it as a finger forgery attack.

[0017] Based on the OCT volume data, the present invention extracts the inner and outer fingerprints and the position of the subcutaneous sweat glands, which means that the full finger data must be collected to perform the forgery attack detection.

[0018] Furthermore, in step 2), the steps of obtaining the OCT inner and outer fingerprints and the position of the subcutaneous sweat glands are not limited to a single acquisition algorithm. Simple or complex OCT inner and outer fingerprint and subcutaneous sweat gland extraction methods and different minutiae extraction algorithms are applicable to the present forgery attack detection method. Different num1, num2, and num3 can be set in step 4) according to the characteristics of different algorithms to obtain the best effect.

[0019] Still further, in step 5), the fingerprint matching algorithm used can be a commercial fingerprint matching algorithm or other non-commercial fingerprint matching algorithms. Different thresholds t can be set according to different fingerprint matching algorithms to obtain the best effect.

[0020] Still further, in step 9), the formula for calculating the coincidence rate between the position of the subcutaneous sweat glands and the inner fingerprint ridges is

[0021] C = N inridge / N all

[0022] Where C is the overlap rate between the subcutaneous sweat gland position and the inner fingerprint ridge, N inridge is the number of sweat glands on the ridges of the inner fingerprint, N all is the total number of sweat glands detected. According to the physiological structure of the finger, sweat glands should all be located on the inner fingerprint ridges. Therefore, this method can correctly detect people whose outer fingerprints are worn but still retain inner fingerprints and sweat glands, demonstrating high compatibility. The final threshold t can be adaptively set based on the tolerance of the actual usage scenario.

[0023] The technical concept of the present invention is: biometric recognition technology is a technology that uses computers to identify individuals using inherent physiological characteristics (fingerprints, irises, facial features, DNA, etc.) or behavioral characteristics (gait, habits, etc.). Fingerprints, as the most widely used biometric feature, are highly vulnerable to counterfeiting attacks. According to physiological information, the inner fingerprint and outer surface fingerprint of the same finger share the same ridges. Furthermore, the spatial structure of sweat glands, which serve as channels for sweat discharge between the epidermis and dermis, indicates that sweat glands are only present within the ridges of the inner and outer fingerprints. Currently known counterfeiting methods primarily focus on counterfeiting outer fingerprints. This is because counterfeiting internal information is difficult, and existing commercial fingerprint collectors can only capture the ridges of outer fingerprints and cannot obtain internal information, making counterfeiting internal fingerprint ridges and sweat gland information extremely difficult. For some people whose outer fingerprints are difficult to identify, the overlap between the ridges of the inner fingerprint and the sweat glands can be exploited to prevent counterfeiting attacks while preserving the identifiable inner fingerprint information. Combining these characteristics, the present method achieves a highly accurate fingerprint counterfeiting attack detection method.

[0024] The present invention's method for detecting finger forgery attacks based on OCT volume data fully utilizes finger OCT volume data and extracts internal and external information to accurately detect forgery attacks, addressing the inability of conventional fingerprint systems to resist forgery attacks. This forgery attack detection process is highly compatible with various OCT fingerprint acquisition systems, ensuring detection of both healthy and worn fingers. It can serve as a pre-processing method for fingerprint recognition systems to protect against various types of forgery attacks.

[0025] This invention fully utilizes finger OCT volume data and extracts internal and external information to accurately detect forgery attacks, addressing the inability of conventional fingerprint systems to resist forgery attacks. This invention is applicable to different OCT fingerprint acquisition systems and offers high compatibility, ensuring detection of both healthy and worn fingers. It can serve as a pre-processing method for fingerprint recognition systems, defending against various forgery attacks.

[0026] The beneficial effects of the present invention are mainly manifested in: accurate forgery attack detection, strong adaptability, and extremely high security and versatility. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is the flow chart of the method of the present invention. Specific implementation

[0028] The following is a more detailed description in conjunction with the accompanying drawings:

[0029] Refer to Figure 1 , a method for detecting finger forgery attacks based on OCT volume data, the process is as follows:

[0030] A method for detecting finger forgery attacks based on OCT volume data, the method includes the following steps:

[0031] 1) Use an OCT fingerprint acquisition device to obtain the OCT volume data of the finger;

[0032] 2) Obtain external fingerprint images, internal fingerprint images, and subcutaneous sweat gland images respectively;

[0033] 3) Detect the number of minutiae of the internal and external fingerprints and the number of subcutaneous sweat glands respectively;

[0034] 4) Set the minutiae number thresholds num1, num2, the sweat gland number threshold num3, the internal and external fingerprint matching score threshold t, and the coincidence rate n of the subcutaneous sweat gland position and the internal fingerprint ridge line;

[0035] 5) If the number of minutiae of the internal fingerprint is more than num1 and the number of minutiae of the external fingerprint is more than num2, calculate the internal and external fingerprint matching score through a fingerprint matcher; if the number of minutiae of the internal fingerprint is less than num1, directly determine it as a finger forgery attack.

[0036] 6) If the number of sweat glands is less than num3, directly determine it as a finger forgery attack;

[0037] 7) If the number of sweat glands is more than num3, determine whether the internal and external fingerprint matching score obtained in step 5) is higher than the set threshold t. If it is higher than the threshold t, the system directly passes the detection. If it is lower than the threshold t, enter step 9);

[0038] 8) If the number of sweat glands is less than num3 and at the same time the number of minutiae of the internal fingerprint is less than num1, enter step 9);

[0039] 9) Locate the position of the subcutaneous sweat gland and calculate the coincidence rate of the subcutaneous sweat gland position and the ridge line of the internal fingerprint;

[0040] 10) If the coincidence rate is higher than the set value n, pass the detection, otherwise determine it as a finger forgery attack.

[0041] This method extracts the internal and external fingerprints and the positions of subcutaneous sweat glands based on OCT volume data, which means that the OCT data of the entire finger must be collected to detect forgery attacks. Since the sweat glands are relatively small, the lateral resolution of the OCT system should be no less than 1000 dpi.

[0042] In step 2), the methods for obtaining the external fingerprint image, internal fingerprint image, and subcutaneous sweat gland image are not limited to a certain algorithm. Simple or complex OCT internal and external fingerprint and subcutaneous sweat gland extraction methods, as well as different minutiae extraction algorithms, are applicable to this forgery attack detection method. Different num1, num2, and num3 can be set in step 4) according to the characteristics of different algorithms to obtain the best results.

[0043] In step 5), the fingerprint matching algorithm used can be a commercial fingerprint matching algorithm or other non-commercial fingerprint matching algorithms. Different thresholds t can be set according to different fingerprint matching algorithms to obtain the best results.

[0044] In step 9), the calculation formula for the coincidence rate between the position of the subcutaneous sweat gland and the ridge line of the internal fingerprint is

[0045] C = N inridge / N all

[0046] where C is the coincidence rate between the position of the subcutaneous sweat gland and the ridge line of the internal fingerprint, and N inridge is the number of sweat glands on the ridge line of the internal fingerprint, and N all is the total number of detected sweat glands. According to the physiological structure of the finger, the sweat glands should all be on the ridge line of the internal fingerprint. Therefore, this method can correctly detect people with worn external fingerprints but still retaining internal fingerprints and sweat glands, with high compatibility. The value of the final threshold t can be adaptively set according to the tolerance of the actual usage scenario.

[0047] This method has 5 thresholds, namely the minutiae number thresholds num1 and num2, the sweat gland number threshold num3, the internal and external fingerprint matching score threshold t, and the coincidence rate n between the position of the subcutaneous sweat gland and the ridge line of the internal fingerprint. In the actual deployment of different fingerprint OCT systems, only a few fingers need to be collected to obtain positive samples, and a few forged samples are collected as negative samples. Then, the 5 thresholds in this method can be determined according to the maximum inter-class variance method or the maximum entropy threshold segmentation method.

[0048] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited by the described embodiments. Any other modifications, substitutions, combinations, and cuttings made without departing from the spirit and principle of the present invention shall be equivalent replacement methods and are all included in the protection scope of the present invention.

Claims

1. A method for detecting finger forgery attacks based on OCT volume data, characterized in that: The method includes the following steps: 1) Use an OCT fingerprint acquisition device to obtain the OCT volume data of the finger; 2) Obtain external fingerprint images, internal fingerprint images, and subcutaneous sweat gland images respectively; 3) Detect the number of minutiae points of the internal and external fingerprints and the number of subcutaneous sweat glands respectively; 4) Set the minutiae point number thresholds num1, num2, the sweat gland number threshold num3, the internal and external fingerprint matching score threshold t, and the coincidence rate n of the subcutaneous sweat gland position and the internal fingerprint ridge line; 5) If the number of internal fingerprint minutiae points is more than num1 and the number of external fingerprint minutiae points is more than num2, calculate the internal and external fingerprint matching score through a fingerprint matcher; if the number of internal fingerprint minutiae points is less than num1, directly judge it as a finger forgery attack; 6) If the number of sweat glands is less than num3, directly judge it as a finger forgery attack; 7) If the number of sweat glands is more than num3, judge whether the internal and external fingerprint matching score obtained in step 5) is higher than the set threshold t. If it is higher than the threshold t, the system directly passes the detection. If it is lower than the threshold t, enter step 9); 8) If the number of sweat glands is less than num3 and at the same time the number of internal fingerprint minutiae points is less than num1, enter step 9); 9) Locate the position of the subcutaneous sweat gland and calculate the coincidence rate of the subcutaneous sweat gland position and the internal fingerprint ridge line; the calculation formula for the coincidence rate of the subcutaneous sweat gland position and the internal fingerprint ridge line is C=N inridge N all where C is the coincidence rate between the position of the subcutaneous sweat glands and the ridges of the inner fingerprint, N inridge is the number of sweat glands on the ridges of the inner fingerprint, N all is the total number of detected sweat glands; according to the physiological structure of the finger, the sweat glands should all be on the ridges of the inner fingerprint. Therefore, it is possible to correctly detect people with worn outer fingerprints but still retaining the inner fingerprint and sweat glands, with high compatibility; the value of the final threshold t can be adaptively set according to the tolerance of the actual usage scenario; 10) If the coincidence rate is higher than the set value n, pass the detection, otherwise judge it as a finger forgery attack.

2. The method for detecting finger forgery attacks based on OCT volume data according to claim 1, characterized in that: Extract the internal and external fingerprints and the subcutaneous sweat gland position based on the OCT volume data, which means that the entire finger data must be collected to perform forgery attack detection.

3. The method for detecting finger forgery attacks based on OCT volume data according to claim 1, wherein: In step 2), the steps of obtaining the OCT internal and external fingerprints and the subcutaneous sweat gland position are not limited to a single acquisition algorithm. Simple or complex OCT internal and external fingerprint and subcutaneous sweat gland extraction methods and different minutiae point extraction algorithms are applicable to this forgery attack detection method. Different num1, num2, and num3 can be set in step 4) according to the characteristics of different algorithms to obtain the best effect.

4. The method for detecting finger forgery attacks based on OCT volume data according to claim 1, characterized in that: In step 5), the fingerprint matching algorithm used can be a commercial fingerprint matching algorithm or other non-commercial fingerprint matching algorithms. Different thresholds t can be set according to different fingerprint matching algorithms to obtain the best effect.

5. The method for detecting finger forgery attacks based on OCT volume data according to claim 1, wherein: Fully consider the unique and permanent modal features in the finger, that is, ensure the detection of normal fingers and the detection of fingers with epidermal wear. It can be used as a preprocessing method for fingerprint recognition systems to resist different types of forgery attacks and has strong anti-counterfeiting performance.

Images